What problem does it solve? AI-generated investigation and bug reports often assert negative conclusions like "the column does not exist" or "no data was found" based only on code reading, leading to wrong decisions. This Skill requires every negative claim to be backed by actual command output or query results. ## Core Features & Use Cases - Mandatory Evidence for Negative Claims: Maps each claim type (missing DB column, missing data, missing function, missing config value) to the exact verification command required, such as SHOW COLUMNS, SELECT COUNT(*), or grep. - External Data Investigation Rules: Requires inspecting the full structure of external tables and third-party data sources before drawing conclusions, since column names and value ranges are unpredictable. - Report Template and Anti-Hallucination Checklist: Provides a structured report format (symptoms, investigation steps, root cause, evidence, fix plan, verification) plus a checklist covering types, foreign keys, soft deletes, and environment differences. - Use Case: When writing a bug report claiming a database column is missing, run SHOW COLUMNS first and paste the actual output into the report instead of relying on code reading. ## Quick Start Ask the AI to write an investigation report for a bug or data issue and require evidence for every negative conclusion it makes.
